Text Generation
Transformers
PyTorch
TensorBoard
Safetensors
bloom
Eval Results
text-generation-inference
Inference Endpoints

AttributeError

#239
by GaaraOtheSand - opened

AttributeError: 'BloomForCausalLM' object has no attribute 'word_embeddings'. I'm actually not too positive what the Accelerate program is trying to do but when I run my program I'm getting this error. Any ideas on how to fix this would be greatly appreciated.

I'm also getting an issue of ValueError: word_embeddings.weight doesn't have any device set. The first checkpoint seems to be too large to load into my text editor so I can't exactly look at the file to edit it, if it even can be.

sounds like someone needs to check out the plugins section of notepad++ ✌️🫡📝 . but... typically... that not usually ever what's wrong unless corrupted download, which well, only one way to fix that anyway. It's typically your settings. sometimes models get reaaaal picky and only 1 way they like to load. I'm a Webui guy though so, idk guess the pipeline for diffusers folk is supposed to be automatic? try googling some settings or just fiddling with different decoders and loaders is what works for me.
And if size were the only issue, it would be a memory error or Cuda error.

christopher changed discussion status to closed

Sign up or log in to comment